JP2012530978A - Method and apparatus for performing shift and exclusive OR operations with a single instruction - Google Patents

Abstract

A method and apparatus for performing shift and XOR operations is disclosed. In one embodiment, the apparatus includes an execution resource that executes the first instruction. In response to the first instruction, the execution resource shifts and XORs to at least one value.
[Selection] Figure 1A

Description

  The present disclosure relates to the computer processing field. More particularly, embodiments relate to instructions that perform shift and logical OR (XOR) operations.

  Single instruction multiple data processing (SIMD) instructions are useful in a variety of applications that process many data elements (packed data) in parallel. Performance decreases when operations such as shift operations or exclusive OR (XOR) operations are executed in series.

  The present invention is illustrated by way of example and not limitation in the accompanying drawings.

It is a block diagram of the computer system comprised with the processor containing the execution part which performs the shift and XOR operation instruction in one Embodiment of this invention. It is a block diagram of another example computer system in another embodiment of the present invention. It is a block diagram of another example computer system in another embodiment of the present invention. 1 is a block diagram of a processor micro-architecture of one embodiment including a logic circuit that performs shift and XOR operations in the present invention. FIG. FIG. 6 is a representation of various pack data types in a multimedia register in one embodiment of the invention. Fig. 9 illustrates a pack data type in another embodiment sentence. Fig. 5 shows various signed and unsigned packed data type representations of multimedia registers in one embodiment of the present invention. 3 illustrates one embodiment of an operational code (opcode) format. Another operational code (opcode) forat is shown. Another operational code format is shown. FIG. 3 is a block diagram of one embodiment of logic for executing instructions in the present invention. It is a flowchart of the calculation performed in one Embodiment.

  The following description describes embodiments of techniques for performing shift or XOR operations on a processing device, computer system, or software program. In the following description, numerous details are set forth such as processor type, microarchitecture requirements, events, enablement mechanism, etc. to facilitate a more complete understanding of the present invention. However, it will be apparent to one skilled in the art that embodiments of the invention may be practiced without these specific details. In addition, well-known structures, circuits, etc. are avoided in detail so as not to unduly obscure the embodiments of the present invention.

  Although the following embodiments are described in terms of a processor, other embodiments are applicable to other types of integrated circuits and logic devices. The same techniques and teachings of the present invention can be readily applied to other types of circuits or semiconductor devices that benefit from increased pipeline throughput and performance. The teachings of the present invention are applicable to any processor or machine that performs data manipulation. However, embodiments of the present invention are not limited to processors or machines that perform 256-bit, 128-bit, 64-bit, 32-bit, or 16-bit data processing, and can be applied to any processor and machine that requires manipulation of packed data. Applicable.

  The following examples describe instruction processing and distribution in the context of execution units and logic circuits, but other embodiments of the invention can be performed by software stored on tangible media. In one embodiment, the method of the present invention is embodied in machine-executable instructions. The instructions can be used to cause a general purpose processor or a dedicated processor programmed with instructions to perform the steps of the present invention. Embodiments of the present invention can be provided as a computer program product or software that can include a machine or computer readable medium that stores instructions that can be used to program a computer (other electronic device) to perform the processes of the present invention. is there. Alternatively, each stage of the invention may be performed by a special purpose hardware component that includes the hard wire logic that performs each stage, or by any combination of programmed computer components and custom hardware components. These software can be stored in the memory of the system. Similarly, the code can be distributed over other computer-readable media or over a network.

  Machine-readable media may include any mechanism for storing or transmitting information in a machine-readable format, including but not limited to floppy disks, optical disks, CDs, CD-ROMs, and magnetic optical disks, ROM, RAM, EPROM, EEPROM, optical card or magnetic card, flash memory, transmission via the Internet, electricity, light, sound, and other forms of propagation signals (eg, carrier wave, infrared signal, digital signal, etc.) are included. Accordingly, computer readable media include any type of media / opportunity or read media suitable for storing or transmitting electronic instructions or information in a form readable by a machine (computer). Furthermore, the present invention can also be downloaded as a computer program product. Thus, the program may be transferred to a requesting computer (eg, client) from a remote computer (eg, server). The transfer of the program can be performed by electrical, optical, acoustic, or other form of data signal embodied on a carrier wave or other propagation medium via a communication link (eg, modem, network connection, etc.).

  Design can be done through various stages from creation to simulation, and thus manufacturing. Data representing the design may represent the design in multiple ways. First, although suitable for simulation, the hardware can be expressed in a hardware description language or other function description language. Furthermore, circuit level models with logic and / or transistor gates can be produced at several stages of the design process. In addition, most designs reach a data level that represents the physical location of various devices in the hardware model at some stage. When using normal semiconductor manufacturing technology, the data representing the hardware model is data indicating the presence or absence of various features on various mask layers for masks used in the manufacture of integrated circuits. Good. In any representation of the design, the data may be stored on any form of machine-readable medium. Magneto-optical storage such as light waves or radio waves, memory, or disks that are modulated or generated to transmit this information is a machine-readable medium. Any of these media may be capable of “carrying” or “indicating” design or software information. When an electrical carrier that directs or carries code or design is transmitted, a new copy is generated in the scope of copying, buffering, or retransmitting the electrical signal. Accordingly, a communication provider or network provider can produce a copy of an article (carrier wave) that embodies the techniques of the present invention.

  Modern processors use a plurality of different execution units to process and execute various codes and instructions. Some are quick to complete, others require a huge number of clock cycles, and not all instructions are manufactured in the same way. Fast instruction throughput improves overall processor performance. Therefore, it is preferable that many instructions are executed at higher speed. However, some instructions are high in complexity and require more execution time and processor resources. Examples include floating point instructions, load / store processing, data movement, and the like.

  As the number of computer systems utilized in the Internet and multimedia applications has increased, the introduction of additional processor support has increased. For example, single instruction multiple data (SIMD) integer / floating point instructions and streaming SIMD extension (SSE) are instructions that reduce the total number of instructions required to perform a particular program task and reduce power consumption. Can do. This type of instruction can speed up software performance by performing parallel processing on multiple data elements. This results in performance gains in a wide range of applications including video, audio, and image / photo processing. There are usually many challenges to implementing SIMD instructions in microprocessors and similar types of logic circuits. Furthermore, SIMD operations are complex and often require additional circuitry to accurately process and manipulate the data.

  At present, SIMD shift and XOR instructions have not been put to practical use. In embodiments of the present invention, without SIMD shift and XOR instructions, multiple instructions and data registers are required to achieve equivalent results in applications such as audio / video / graphics compression, processing, manipulation, etc. Is assumed. Accordingly, code overhead and resource requirements can be reduced by utilizing at least one shift and XOR instruction in embodiments of the present invention. Embodiments of the present invention provide a method for implementing shift and XOR operations as algorithms that utilize SIMD related hardware. Currently, it is difficult and time consuming to perform shift and XOR operations on data in SIMD registers. Some algorithms have more instructions for allocating arithmetic operation data than the actual number of instructions required to execute these operations. By implementing the shift and XOR operations in the embodiments of the present invention, the number of instructions required to perform the shift and XOR operations can be greatly reduced.

  Embodiments of the invention relate to implementation instructions for shift and XOR operations. In one embodiment, the shift and XOR operations are ...

  The shift and XOR operations in one embodiment performed on the data elements can be generally expressed as DEST1 ← SCR1 “SRC2”.

  In one embodiment, SRC1 stores a first operand having a plurality of data elements, and SRC2 includes a value representing a value that is shifted with a shift and XOR instruction. In other embodiments, shift and XOR value indicators may be stored in the immediate field.

  In the flow described above, “DEST” and “SRC” are general terms that represent the source and destination of the corresponding data or operation. In some embodiments, these can be implemented in registers, memory, or other storage areas having different names or functions than those described. In one embodiment, for example, DEST1 and DEST2 may be a first time storage area and a second time storage area (eg, “TEMP1” and “TEMP2” registers), and SRC1 and SRC3 are first and second Destination storage area (for example, “DEST1” and “DEST2” registers). In other embodiments, two or more SRC and DEST storage areas may correspond to different data storage elements within the same storage area (eg, a SIMD register).

  FIG. 1A is a block diagram of a computer system formed by a processor including an execution unit that executes shift and XOR operation instructions according to an embodiment of the present invention. The system 100 includes components, such as a processor 102 that utilizes an execution unit that includes logic to execute an algorithm for processing data in the present invention, such as in the embodiments described herein. System 100 may be PENTIUM® III, PENTIUM® 4, Xeon®, Itanium®, XScale®, and / or available from Intel Corporation of Santa Clara, California. However, other systems (including other microprocessors, PCs with engineering workstations, set-top boxes, etc.) may be utilized. In one embodiment, the sample system 100 can run a version of the WINDOWS® operating system available from Microsoft Corporation, Redmond, W., but other operating systems (eg, UNIX®). ), Linux), embedded software, and / or a graphic user interface. Thus, embodiments of the invention are not limited to a specific combination of hardware circuitry and software.

  Embodiments are not limited to computer systems. Other embodiments of the invention can be utilized for other devices such as handheld devices and embedded applications. Examples of handheld devices include cellular phones, internet protocol devices, digital cameras, personal digital assistants (PDAs), and handheld PCs. Embedded applications can be microcontrollers, digital signal processors (DSPs), system-on-chips, network computers (NetPCs), set-top boxes, network hubs, wide area network (WAN) switches, or performing shift and XOR operations on operands Any other possible system can be included. In addition, to increase the efficiency of multimedia applications, several architectures can be implemented to execute instructions on several data simultaneously. As the type and amount of data increases, it becomes necessary to improve the computer and its processor to manipulate the data in a more efficient manner.

  FIG. 1A is a block diagram of a computer system 100 comprised of a processor 102 that includes one or more execution units 108 that execute an algorithm that shifts and XORs several data elements in one embodiment of the invention. One embodiment is described assuming a single processor desktop or server system, but may include other embodiments that assume multiple processors. System 100 is an example of a hub architecture. Computer system 100 includes a processor 102 that processes data signals. The processor 102 may be a complex instruction set computer (CISC) microprocessor, a reduced instruction set computer (RISC) microprocessor, a very long instruction (VLIW) microprocessor, a processor implementing a combination of instruction sets, or any other processor device. (For example, a digital signal processor). The processor 102 is coupled to a processor bus 110 that can transmit data signals between the processor 102 and other processors in the system 100. The elements of system 100 perform their normal functions known to those skilled in the art.

  In one embodiment, the processor 102 includes a level 1 (L1) internal cache memory 104. Depending on the architecture, the processor 102 may have a single internal cache or multiple levels of internal cache. In another embodiment, the cache memory may reside external to the processor 102. As another embodiment, a combination of both an internal cache and an external cache may be included depending on the implementation example and necessity. Register file 106 can store various types of data in various registers, such as integer registers, floating point registers, status registers, and instruction pointer registers.

  Execution unit 108 includes logic that performs integer and floating point operations, also residing within processor 102. The processor 102 further includes a microcode (ucode) ROM that stores microcode for certain macro instructions. In this embodiment, execution unit 108 includes logic to process packed instruction set 109. In one embodiment, packed instruction set 109 includes packed shift and XOR instructions that shift and XOR multiple operands. By including the packed instruction set 109 in the instruction set of the general-purpose processor 102 and further including related circuits for executing the instructions, processing used by many multimedia applications can be performed using the filling data of the general-purpose processor 102. Can do. Thus, many multimedia applications can be accelerated and run more efficiently by utilizing the full width of the processor's data bus to process the fill data. This eliminates the need to transfer each piece of data to the processor data bus when one process is performed on one data element at a time.

  Furthermore, other embodiments of the execution unit 108 can be used in a microcontroller, an embedded processor, a graphic device, a DSP, and other logic circuits. System 100 includes a memory 120. The memory 120 may be a DRAM device, SRAM device, flash memory device, or other memory device. Memory 120 may store instructions and / or data represented by data signals that are executable by processor 102.

  A system logic chip 116 is coupled to the processor bus 110 and the memory 120. The system logic chip 116 in the illustrated embodiment is a memory controller hub (MCH). The processor 102 can communicate with the MCH 116 via the processor bus 110. The MCH 116 provides a high bandwidth memory path 118 to the memory 120 for storing instructions and data and storing graphic commands, data and textures. The MCH 116 transmits data signals between the processor 102, memory 120, and other components of the system 100 to bridge the data signals between the processor bus 110, memory 120, and system I / O 122. In some embodiments, the system logic chip 116 may provide a graphics port for coupling to the graphics controller 112. The MCH 116 is coupled to the memory 120 via the memory interface 118. The graphics card 112 is coupled to the MCH 116 via an accelerated graphics port (AGP) interconnect 114.

  The system 100 utilizes a dedicated hub interface bus 122 to connect the MCH 116 to an I / O controller hub (ICH) 130. The ICH 130 provides a direct connection via a local I / O bus to some I / O devices. The local I / O bus is a high-speed I / O bus for connecting peripheral devices to the memory 120, the chipset, and the processor 102. Some examples include audio controllers, firmware hubs (flash BIOS) 128, wireless transceivers 126, data storage 124, legacy I / O controllers including user input and keyboard interfaces, serial expansion ports such as Universal Serial Bus (USB), And a network controller 134. The data storage device 124 may include a hard disk drive, floppy disk drive, CD-ROM device, flash memory device, or other mass storage device.

  In another embodiment of the system, an execution unit that executes an algorithm with shift and XOR instructions can be utilized with the system on chip. One embodiment of a system on chip consists of a processor and memory. An example of such a system memory is a flash memory. The flash memory may be located on the same die as the processor and other system components. In addition, other logic blocks (eg, memory controller or graphics controller) can be located on the system on chip.

  FIG. 1B illustrates a data processing system 140 that implements the principles of one embodiment of the invention. Those skilled in the art will appreciate that the embodiments described herein may be applied to other processing systems without departing from the scope of the present invention.

  Computer system 140 includes a processing core 159 that has the capability of performing SIMD operations including shift and XOR operations. In one embodiment, processing core 159 represents a processing unit of any type of architecture (such as, but not limited to, a CISC, RISC, or VLIW architecture). The processing core 159 may be suitable for manufacturing with one or more process technologies, or may be suitable for facilitating this manufacturing by representing sufficient details in a machine-readable medium. .

  The processing core 159 includes an execution unit 142, a register file set 145, and a decoder 144. Processing core 159 further includes circuitry (not shown) that is not necessary for an understanding of the present invention. The execution unit 142 is used to execute a command received by the processing core 159. In addition to recognizing normal processor instructions, the execution unit 142 can recognize instructions in the pack instruction set 143 and perform operations on the pack data format. The packed instruction set 143 includes instructions that support shift and XOR operations, and may also include other packed instructions. The execution unit 142 is connected to the register file 145 by an internal bus. The register file 145 represents a storage area on the processing core 159 that stores information including data. As described above, a storage area used for storing pack data is not essential. The execution unit 142 is connected to the decoder 144. The decoder 144 is used to decode instructions received by the processing core 159 into control signals and / or microcode entry points. In response to these control signals and / or microcode entry points, the execution unit 142 performs appropriate processing.

  The processing core 159 is coupled to a bus 141 that communicates with various other system devices including, but not limited to, SDRAM control 146, SRAM control 147, burst flash memory interface 148, PCMCIA (personal A computer memory card international association / compact flash (CF) card control 149, a liquid crystal display (LCD) control 150, a direct memory access (DMA) controller 151, and a substitute bus master interface 152 are included. In one embodiment, the data processing system 140 may further include an I / O bridge 154 that communicates with various I / O devices via the I / O bus 153. The I / O device further includes, but is not limited to, a universal asynchronous receiver / transmitter (UART) 155, a universal serial bus (USB) 156, a Bluetooth® wireless UART 157, and an I / O expansion interface 158. It's okay.

  One embodiment of the data processing system 140 provides a processing core 159 that provides mobile, network and / or wireless communication and has the capability to perform SIMD operations including shift and XOR operations. The processing core 159 includes various audio, video, image and communication algorithms (eg, Walsh Hadamard transform, fast Fourier transform (FFT), discrete cosine transform (DCT), and their respective inverse transforms, color space transforms, etc. It may be programmed with compression / decompression techniques, video coding motion estimation or video decoding motion compensation, and modulation and demodulation (MODEM functions such as pulse code modulation (PCM)). Some embodiments of the present invention may also be used for graphic applications (three-dimensional (“3D”) modeling, drawing, object collision detection, 3D object conversion, lighting, etc.).

  FIG. 1C illustrates another embodiment of a data processing system having the capability of performing SIMD shift and XOR operations. In another embodiment, the data processing system 160 may include a main processor 166, a SIMD coprocessor 161, a cache memory 167, and an input / output system 168. The input / output system 168 is not essential, but may be coupled to the wireless interface 169. The SIMD coprocessor 161 has a function of performing SIMD operations including shift and XOR operations. The processing core 170 may be suitable for manufacturing with one or more process technologies, and may represent all or part of the data processing system 160 including the processing core 170 by representing sufficient details in a machine-readable medium. It may be suitable for promoting the manufacture of

  In one embodiment, the SIMD coprocessor 161 includes an execution unit 162 and a register file set 164. One embodiment of the main processor 165 includes a decoder 165 that recognizes instructions in the instruction set 163, including SIMD shift and XOR calculation instructions executed by the execution unit 162. In another embodiment, the SIMD coprocessor 161 further includes at least a portion of a decoder 165B that decodes the instructions in the instruction set 163. Processing core 170 further includes circuitry (not shown) that is not necessary for an understanding of the present invention.

  In operation, main processor 166 executes a data processing instruction stream that controls general types of data processing, including interaction with cache memory 167 and input / output system 168. A SIMD coprocessor instruction is embedded in the data processing instruction stream. The decoder 165 of the main processor 166 recognizes these SIMD coprocessor instructions as being of a type to be executed by the connected SIMD coprocessor 161. Accordingly, the main processor 166 issues these SIMD coprocessor instructions (or control signals representing SIMD coprocessor instructions) on the coprocessor bus 166, from which any connected SIMD coprocessor receives these instructions. . In this case, the SIMD coprocessor 161 accepts and executes all received SIMD coprocessor instructions destined for it.

  Data is received via the wireless interface 169 and prepared for processing by SIMD coprocessor instructions. As an example, voice communications may be received in the form of a digital signal, and upon processing of SIMD coprocessor instructions, digital audio samples representing the voice communications are played. In another example, compressed audio and / or video may be received in the form of a digital bitstream that is processed by SIMD coprocessor instructions to play digital audio samples and / or motion video frames. Good. In one embodiment of processing core 170, main processor 166 and SIMD coprocessor 161 include a single decoder 165 that recognizes instructions in execution unit 162, register file set 164, and instruction set 163 including SIMD shift and XOR instructions. Integrated into the processing core 170.

  FIG. 2 is a block diagram of a micro-architecture of processor 200 of one embodiment that includes logic circuitry for performing shift and XOR operations in the present invention. In one embodiment of the shift and XOR instruction, the instruction shifts the floating point mantissa to the right by the amount indicated by the exponent and XORs the shifted value with the given value to produce the final result. To do. In one embodiment, the normal front end 201 is the part of the processor 200 that fetches the macro instructions to execute and prepares them for use in the processor pipeline. The front end 201 may include several units. In one embodiment, an instruction prefetcher 226 fetches macro instructions from memory and supplies them to an instruction decoder 228, which is capable of executing them by primitives (called microinstructions or microoptions) that can be executed by a machine. (Sometimes referred to as micro-op or u-op). In one embodiment, the trace cache 230 takes the decoded u-ops and assembles them for execution into a sequence or trace that is programmed by the u-op queue 234 program. When the trace cache 230 finds a composite microinstruction, the microcode ROM 232 provides the u option necessary to complete the operation.

  Many macro instructions are converted into a single micro-op, and other macro instructions require several micro-ops to complete the entire operation. In one embodiment, if more than four microops are required to complete a macro instruction, decoder 228 accesses microcode ROM 232 to execute the macro instruction. In one embodiment, the pack shift and XOR instructions are decoded into a small number of microops for preparation at the instruction decoder 228. In another embodiment, instructions for pack shift and XOR algorithms can be stored in the microcode ROM 232 if several numbers of microops are required to perform the processing. The trace cache 230 references the entry point programmable logic array (PLA) to determine the correct microinstruction pointer that reads the microcode sequence for the shift and XOR algorithm of the microcode ROM 232. When the microcode ROM 232 finishes ordering the microops for the current macroinstruction, the machine front end 201 resumes fetching the microops from the trace cache 230.

  SIMD and other multimedia type instructions are considered compound instructions. Most floating point instructions are also compound instructions. Thus, when the instruction decoder 228 finds a composite macro instruction, it accesses the appropriate location in the microcode ROM 232 to obtain the microcode sequence for that macro instruction. Various micro-ops necessary to execute the macro instruction are communicated to the out-of-order execution engine 203 to prepare for execution in the appropriate integer and floating point execution units.

  The out-of-order execution engine 203 prepares micro instructions for execution. Out-of-order execution logic smoothes and reorders the flow of microinstructions to optimize performance as it flows through the pipeline and prepare for execution. The allocator logic allocates machine buffers and resources needed for each u option to execute. Register rename logic renames a logical register to an entry in a register file. The allocator further forwards each u-op entry to either of the two u-op queues before the instruction scheduler, memory scheduler, fast scheduler 202, slow / general floating-point scheduler 204, and simple floating-point scheduler 206. One is assigned to memory processing, and one is assigned to non-memory processing. The u-op scheduler 202, 204, 206 is ready for the u-op based on whether the dependent input register operand source is ready and the availability of execution resources required for the u-op to complete processing. Judgment is made. The fast scheduler 202 of this embodiment schedules each half of the main clock cycle, but other schedulers can schedule only once for each main processor clock cycle. The scheduler coordinates between shipping ports to schedule u-options to execute.

  The register files 208 and 210 exist between the schedulers 202, 204, and 206 and the execution units 212, 214, 216, 218, 220, 222, and 224 (execution block 211). Separate register files 208, 210 exist for integer and floating point operations, respectively. Each register file 208, 210 of this embodiment further includes a bypass network that can now bypass or forward results that have just been completed and have not yet been written to the register file to a new dependent u option. The integer register file 208 and the floating point register file 210 further have a function of communicating data with the other. In one embodiment, the integer register file 208 is split into two separate register files (one register file for the lower 32 bits of data and the other register file for the upper 32 bits of data). Is done. Since floating point instructions are typically 64 to 128 bits wide, the floating point register file 210 of one embodiment has 128 bit wide entries.

  The execution block 211 includes execution units 212, 214, 216, 218, 220, 222, and 224 in which instructions are actually executed. This section includes register files 208, 210 that store the values of integer and floating point data operands that the microinstruction needs to execute. The processor 200 according to the present embodiment includes several execution units (address generation unit (AGU) 212, AGU 214, high-speed ALU 216, high-speed ALU 218, slow ALU 220, floating point ALU 222, floating point moving unit 224). In this embodiment, floating point execution blocks 222, 224 perform floating point MMX, SIMD, and SSE operations. The floating-point ALU 222 of this embodiment includes a 64-bit × 64-bit floating-point divider that calculates micro-op division, square root, and remainder. In embodiments of the present invention, any processing related to floating point values is performed in floating point hardware. For example, a floating point register file is used for conversion between the integer format and the floating point format. Similarly, floating-point division is performed by a floating-point divider. On the other hand, non-floating point and integer types are handled with integer hardware resources. This simple, very frequently performed ALU operation is sent to the high-speed ALU item sections 216, 218. The high-speed ALUs 216 and 218 of this embodiment can perform high-speed processing with an effective latency of one-half clock cycle. In one embodiment, most complex integer operations are sent to the slow ALU 220, which slows the integer execution hardware for long latency types of operations such as multiplication, shift, flag logic, and branch processing. It is because it contains. Memory load / store operations are performed by the AGUs 212 and 214. In the present embodiment, the integer ALUs 216, 218, and 220 are described by taking an integer operation for a 64-bit operand as an example. However, in other embodiments, the ALUs 216, 218, 220 may be implemented to support various data bits such as 16, 32, 128, 256, etc. Similarly, the floating point portions 222, 224 may be implemented to support a range of operands having various width bits. In one embodiment, floating point 222.224 can operate on 128-bit wide packed data operands in conjunction with SIMD and multimedia instructions.

  The term “register” is used herein to indicate an on-board processor storage location that is used as part of a macro instruction that identifies an operand. In other words, the registers used here are those that can be seen from outside the processor (for example, those that are visible to the programmer). However, the meaning of the register in one embodiment is not limited to a particular type of circuit. The registers in one embodiment are only required to be able to store and provide data and to perform the functions described herein. The registers described here can be implemented by circuitry within the processor using any number of different technologies (eg, dedicated physical registers, physical registers dynamically allocated using register renaming functions). , Combinations of dedicated registers and dynamically allocated physical registers). In one embodiment, the integer register stores 32-bit integer data. The register file of one embodiment further includes 16 XMM and general purpose registers, and 8 multimedia (eg, “EM64T” additional) multimedia SIMD registers for packed data. In the following description, the registers are 64-bit wide MMX® registers (sometimes referred to as “mm” registers) in a microprocessor enabled by MMX technology available from Intel Corporation of Santa Clara, California. Is understood as a data register designed to hold packed data. These MMX registers are available in integer and floating point form and can be processed with packed data elements associated with SIMD and SSE instructions. Similarly, a 128-bit wide XMM register for SSE2, SSE3, SSE4, or beyond (generally referred to as “SSEx”) can also be used to hold these packed data operands. In this embodiment, when storing packed data and integer data, the register does not need to distinguish between the two data types. In one embodiment, other registers and register combinations may be utilized to store more than 256 bits of data.

  In the example of the following figure, a plurality of data operands are described. FIG. 3A illustrates various types of pack data representations in multimedia registers in one embodiment of the invention. FIG. 3A shows the data types of packed byte 310, packed word 320, and packed double word (dword) 330 for a 128-bit wide operand. The packed byte format 310 in this example has a length of 128 bits and includes 16 packed byte data elements. One byte is defined as 8-bit data. The information of each byte data element includes bit 7 to bit 0 as byte 0, bit 15 to bit 8 as byte 1, bit 23 to bit 16 as byte 2, and finally bit 120 to bit 127. It is stored as byte 15 and so on. In this way, all available bits can be used in the register. This storage configuration increases the storage efficiency of the processor. Further, when accessing 16 data elements, one process can be executed in parallel for 16 data elements.

  In general, one data element refers to individual data stored in a single register or memory location along with other data elements of the same length. In the packed data sequence related to the SSEx technology, the number of data elements stored in the XMM register is a value obtained by dividing 128 bits by the bit length of each data element. Similarly, in packed data sequences for MMX and SSE techniques, the number of data elements stored in the MMX register is a value obtained by dividing 64 bits by the bit length of each data element. Although the type of data shown in FIG. 3A is 128 bits long, embodiments of the present invention can handle operands that are 64 bits wide or of other sizes. The packed word format 320 in this example is 128 bits long and includes eight packed word data elements. Each packed word contains 16 bits of information. The packed double word format 330 of FIG. 3A is 128 bits long and includes four packed double word data elements. Each packed double word data element contains 32 bits of information. The packed quadword is 128 bits long and contains two packed quadword data elements.

  FIG. 3B shows a data storage format in another register. Each pack data may include more than one independent data element. Three pack data elements are described: pack half 341, pack single 342, and pack double 343. One embodiment of packed half 341, packed single 342, and packed double 343 includes fixed-point data elements. In another embodiment, one or more of pack half 341, pack single 342, and pack double 343 include floating point data elements. Another embodiment of the pack half 341 is 128 bits long with eight 16-bit data elements. One embodiment of packed single 342 is 128 bits long and includes four 32-bit data elements. One embodiment of packed double 343 is 128 bits long and includes two 64-bit data elements. These packed data formats can be further extended to other register lengths (eg, 96 bits, 160 bits, 192 bits, 224 bits, 256 bits, or more).

  FIG. 3C shows various signed and unsigned types of packed data representations of multimedia registers in one embodiment of the present invention. Unsigned packed byte representation 344 indicates that unsigned packed bytes are stored in the SIMD register. The information of each byte data element includes bit 7 to bit 0 as byte 0, bit 15 to bit 8 as byte 1, bit 23 to bit 16 as byte 2, and finally from bit 120 to bit 127. Is stored as byte 15 and so on. In this way, all available bits can be used in the register. This storage configuration increases the storage efficiency of the processor. Further, in this configuration, when 16 data elements are accessed, one process can be executed in parallel on the 16 data elements. Signed packed data representation 345 indicates the storage state of signed packed bytes. The eighth bit of each byte data element is a sign indicator. Unsigned packed data representation 346 shows how words 7 through 0 are stored in the SIMD register. Signed packed word representation 347 is similar to representation 346 in the register of unsigned packed words. The 16th bit of each word data element is a sign indicator. Unsigned packed doubleword representation 348 shows how doubleword data elements are stored. Signed packed doubleword representation 349 is similar to representation 348 in the register of unsigned packed doublewords. The required sign bit is the 32nd bit of each doubleword data element.

  FIG. 3D shows an embodiment of an opcode format 360, which has 32 or more bits and has a register / memory operand address mode of “IA-32 Intel Architecture Software Developer's Manual Volume 2: Instruction Set Reference”. This document is published by Intel Corporation of Santa Clara, California, and is available from www.intel.com/design/litcentr. In one embodiment, shift and XOR operations may be encoded with one or more of fields 361 and 362. For each instruction, specify up to two operand positions (including up to two source operand identifiers 364 and 365). In one embodiment of the shift and XOR instruction, the destination operand identifier 366 is equal to the source operand identifier 364, but may be different in other embodiments. In another embodiment, the destination operand identifier 366 is equal to the source operand identifier 365, but may be different in other embodiments. In one embodiment of the shift and XOR instruction, one of the source operands identified by the source operand identifiers 364 and 365 is overwritten with the result of the shift and XOR operand, and in another embodiment, the identifier 364 corresponds to a source register element. The identifier 365 corresponds to the destination register element. In one embodiment of the shift and XOR instructions, operand identifiers 364 and 365 are utilized to identify 32-bit or 64-bit source and destination operands.

  FIG. 3E shows another operational code (opcode) format 370 having 40 or more bits. Opcode format 370 corresponds to opcode format 360 and includes an optional prefix byte 378. This type of shift and XOR operation may be encoded in one or more of fields 378, 371, and 372. Up to two operand positions can be specified for each instruction by source operand identifiers 374 and 375 and by prefix byte 378. In one embodiment of shift and XOR instructions, prefix byte 378 is utilized to identify 32-bit or 64-bit source and destination operands. In one embodiment of the shift and XOR instruction, the destination operand identifier 376 is equal to the source operand identifier 374, but may be different in other embodiments. In another embodiment, destination operand identifier 376 is equal to source operand identifier 375, but may be different in other embodiments. In one embodiment of the shift and XOR instruction, one of the operands identified by operand identifiers 374 and 375 is shifted and XORed to another operand identified by operand identifiers 374 and 375, and this is done with the shift and XOR Overwriting with the result, but in other embodiments, the operand shifts and XORs identified by identifiers 374 and 375 are written to another data element of another register. Opcode formats 360 and 370 are register-to-register, memory-to-register, memory-to-register, register-to-register, immediate-to-register, register-to-register, partially through MOD fields 363 and 373, and optionally scale index Write to a memory address specified by the base and displacement bytes.

  Referring now to FIG. 3F, in some other embodiments, 64-bit single instruction multiple data (SIMD) arithmetic operations may be performed by coprocessor data processing (CDP) instructions. Opcode format 380 shows such a CDP instruction with CDP opcode fields 382 and 389. This type of CDP instruction may be encoded with one or more of fields 383, 384, 387, and 388 in another embodiment of shift and XOR operations. For each instruction, up to three operand positions are identified, including up to two source operand identifiers 385 and 390 and one destination operand identifier 386. One embodiment of a coprocessor can operate on 8, 16, 32, and 64-bit values. In one embodiment, shift and XOR operations are performed on floating point data elements. In some embodiments, shift and XOR instructions may be conditionally executed utilizing selection field 381. For some shift and XOR instructions, the source data size may be encoded by field 383. In some embodiments of the shift and XOR instructions, zero (Z), negative (N), carry (C), and overflow (V) detection is performed on the SIMD field. For some instructions, the type of saturation can also be encoded in field 384.

  FIG. 4 is a block diagram of one embodiment of logic for performing shift and XOR operations on packed data operands in the present invention. Embodiments of the present invention can be implemented in functions having various types of operands such as those described above. For brevity, the following description and examples are described by way of example of shift and XOR instructions that process data elements. In one embodiment, first operand 401 is shifted by shifter 410 by the amount specified by input 405. In one embodiment, this is a right shift. However, in other embodiments, the shifter performs a left shift operation. In some embodiments, the operand is a scalar value, while in other embodiments it is a packed data value (eg, floating point, integer) that has a plurality of different possible data sizes and types. In one embodiment, the shift count 405 is a packed (or “vector”) value, each element corresponding to an element of the packed operand that is shifted by the corresponding shift count element. In other embodiments, the shift count is applied to all elements of the first data operand. Further, in some embodiments, the shift count is specified in one field of the instruction (eg, immediate, r / m, or other field). In other embodiments, the shift count is specified by a register specified by the instruction.

  The shifted operand is then XORed by logic 420 with the value 430 and the XORed result is stored in the destination storage location (eg, register) 425. In one embodiment, the XOR value 430 is a packed (or “vector”) value, each element corresponding to an element of the packed operand that is XORed with the corresponding XOR element. In other embodiments, the XOR value 430 applies to all elements of the first data operand. Further, in some embodiments, the XOR value is specified in one field of the instruction (eg, immediate, r / m, or other field). In other embodiments, the XOR value is specified by a register specified by the instruction.

  FIG. 5 illustrates the operation of shift and XOR instructions in one embodiment of the present invention. When a shift and XOR instruction is received in process 501, the first operand is shifted by the shift count of process 505. In one embodiment, this is a right shift. In other embodiments, the shifter may perform a left shift. In some embodiments, the operand is a scalar value, while in other embodiments it is a packed data value (eg, floating point, integer) that has a plurality of different possible data sizes and types. In one embodiment, the shift count 405 is a packed (or “vector”) value, each element corresponding to an element of the packed operand that is shifted by the corresponding shift count element. In other embodiments, the shift count is applied to all elements of the first data operand. Further, in some embodiments, the shift count is specified in one field of the instruction (eg, immediate, r / m, or other field). In other embodiments, the shift count is specified by a register specified by the instruction.

  In operation 510, the shifted value is XORed with the XOR value. In one embodiment, the XOR value 430 is a packed (or “vector”) value, each element corresponding to an element of the packed operand that is XORed with the corresponding XOR element. In other embodiments, the XOR value 430 applies to all elements of the first data operand. Further, in some embodiments, the XOR value is specified in one field of the instruction (eg, immediate, r / m, or other field). In other embodiments, the XOR value is specified by a register specified by the instruction.

  In operation 515, the shifted and XORed value is stored at a given location. In one embodiment, this location is a scalar register. In another embodiment, this location is a packed data register. In another embodiment, the destination location is also used as the source location (eg, a packed data register specified by the instruction). In other embodiments, the destination location is a different location than the source location storing the first operand or other value (eg, shift count or XOR value).

  In one embodiment, shift and XOR instructions are useful in data de-duplication in various computer applications. Data deduplication is an attempt to find common data blocks between files to optimize disk storage and / or network bandwidth. In one embodiment, the shift and XOR instructions utilize a rolling hash, hash digest (eg, SHA1 or MD5), and native chunk compression (utilizing a fast Rempel-Ziv scheme) to define chunk boundaries. This is useful for improving the deduplication performance of data by using a process such as finding data.

For example, a deduplication algorithm for certain data can be represented by the following pseudo code.

  In the above-described algorithm, the scramble table is a random 32-bit constant 256 entry array, and v is a rolling hash having a hash value of the past 32-byte data. If a chunk boundary is found, the algorithm returns as ret = 1 and position p indicates the chunk boundary. The value z is a value that can detect a good chunk, such as 12 to 15, and may be determined according to the application. In one embodiment, utilizing the shift and XOR instructions, the algorithm described above can be performed at a rate of approximately 2 cycles / byte. In other embodiments, shift and XOR instructions may execute algorithms faster or slower depending on the application.

At least one embodiment that utilizes shift and XOR instructions can be represented by the following pseudo code:

  In the algorithm described above, each entry in the ref1_scramble array includes a version that reflects the bit of the corresponding entry in the original scramble array. In one embodiment, the algorithm described above shifts v to the left rather than to the right, where v includes a version reflecting the bits of the rolling hash. In one embodiment, chunk boundary checking is performed by checking the minimum number of leading zeros.

  In other embodiments, the shift and XOR instructions may be utilized with other useful computer operations and algorithms. Furthermore, the performance of a large number of programs that use shift and XOR operations on a large scale can be improved according to the embodiment.

  Thus, techniques for performing shift and XOR instructions have been disclosed. Although some embodiments are illustrated in the accompanying drawings, these embodiments are intended to be examples only and are not intended to limit the invention to a broader scope, and the present invention may be illustrated or described. It should be noted that the specific configurations and arrangements are not limited. Those of ordinary skill in the art who have read this disclosure will envision various other modifications. Since the art is growing rapidly and it is difficult to foresee future advances, the disclosed embodiments can be made in construction and detail in accordance with the technical advances without departing from the principles of the disclosure or the scope of the appended claims. It should be understood that it can be easily changed.

Claims (36)

  A processor comprising logic to execute a shift and XOR instruction that shifts a first value by a given shift amount and XORs the shifted value with a second value.   The processor of claim 1, wherein the first value is shifted left.   The processor of claim 1, wherein the first value is right shifted.   The processor of claim 1, wherein the first value is logically shifted.   The processor of claim 1, wherein the first value is arithmetically shifted.   The processor according to claim 1, comprising a shifter and an XOR circuit.   The processor of claim 1, wherein the shift and XOR instructions include a first field that stores the second value.   The processor of claim 1, wherein the first value is a packed data type. Storage for storing a first instruction that performs shift and XOR operations;
A processor that executes logic to execute a shift and XOR instruction that shifts a first value by a given shift amount and XORs the shifted value with a second value.   The system of claim 9, wherein the first value is shifted left.   The system of claim 9, wherein the first value is right shifted.   The system of claim 9, wherein the first value is logically shifted.   The system of claim 9, wherein the first value is arithmetically shifted.   The system of claim 9, comprising a shifter and an XOR circuit.   The system of claim 9, wherein the shift and XOR instructions include a first field that stores the second value.   The system of claim 9, wherein the first value is a packed data type.   A method comprising performing a shift and XOR instruction that shifts a first value by a given shift amount and XORs the shifted value with a second value.   The method of claim 17, wherein the first value is shifted left.   The method of claim 17, wherein the first value is right shifted.   The method of claim 17, wherein the first value is logically shifted.   The method of claim 17, wherein the first value is arithmetically shifted.   The method of claim 17, comprising a shifter and an XOR circuit.   The method of claim 17, wherein the shift and XOR instruction includes a first field storing the second value.   The method of claim 17, wherein the first value is a packed data type. A machine-readable medium storing instructions, wherein when the instructions are executed by a machine,
Shifting the first value by a given shift amount;
XORing the shifted value with a second value.   26. The method of claim 25, wherein the first value is shifted left.   26. The method of claim 25, wherein the first value is shifted right.   26. The method of claim 25, wherein the first value is logically shifted.   26. The method of claim 25, wherein the first value is arithmetic shifted.   The method of claim 25, comprising a shifter and an XOR circuit.   26. The method of claim 25, wherein the shift and XOR instructions include a first field that stores the second value.   26. The method of claim 25, wherein the first value is a packed data type. Performing an exclusive OR (XOR) operation between the first shift value and the second bit reflected value, and storing the execution result in the first register;
Checking the minimum number of leading zeros in the execution result.   34. The method of claim 33, wherein the execution result corresponds to a first chunk if the minimum number of leading zeros is in the execution result.   35. The method of claim 34, wherein the first shift value is shifted left by a position corresponding to 1 bit.   The method of claim 34, wherein the first shift value is right shifted by a position corresponding to 1 bit.

Images